Gliomas are the most common primary brain tumors with GBM as the most common type and the most aggressive with median survival of about a year. The GBMs are now divided into 3 or 4 subgroups based on molecular signaling pathways and transcription profiles (Brennan et al, PLOS One 2009). We now have very accurate mouse models of the GBM subgroups and the most widely used is the PDGF-driven model of the proneural subgroup (Dai et al. Genes and Dev 2001). This model has provided us with a great deal of information about the biology of these tumors, especially about the tumor microenvironment around the perivascular niche (Charles et al. Cell Stem Cell 2009). These tumors show a collection of cells around the blood vessels that are not tumor cell-derived but contribute to the biology of the tumor in many ways. These cells include astrocytes, inflammatory cells and smooth muscle cells and provide a haven for tumor cells with stem cell character (Bleau et al. Cell Stem Cell 2009) and provide resistance to therapy (Hambardzumyan et al, Genes and Dev 2008). In this grant we will use novel technology to design mice that will allow us to obtain the translating mRNAs for these various cell types in vivo in these tumors. We will use this technology to measure the expression differences induced by standard of care therapy in an attempt to understand how the various cell types of the perivascular niche contribute to resistance to therapy. Our observafions in the mice will then be compared with human GBM samples of various subtypes to understand the role of these stromal cells of the glioma microenvironment. Aim 1. In mouse models of proneural GBMs we will determine which cells of the perivascular stroma produce the gene products that drive the resistance phenotype in human GBM. Aim 2. In human proneural GBM surgical samples we will verify in which perivascular cell types these genes are produced. Aim 3. In mouse models of proneural GBMs we will determine if expression of these genes in their respective cell types is regulated by aggressiveness of the tumor cells or if it responds to treatment with radiation and temozolomide over time. Aim 4. In mouse models of proneural GBMs we will knock down these genes in the appropriate perivascular cell type to determine if any of them are causally related to therapeufic resistance. Aim 5. In this aim we will extend the above aims to NFI/mesenchymal and EGFR/classical GBMs.